Biomedical Applications of Multi-Material Phase Retrieval in Propagation-Based Phase-Contrast Imaging

Ilian Haggmark, William Vagberg, Hans M. Hertz, Anna Burvall
2018 Microscopy and Microanalysis  
In propagation-based phase-contrast imaging the measured quantity is proportional to the second-order derivative of the phase shift induced by the object. This makes the techniques more sensitive to high spatial frequencies and low-absorbing materials than conventional X-ray imaging. To obtain an image more similar to the object and to get quantitative values it is, however, necessary to retrieve the original phase shift from its second-order derivative. This process, phase retrieval, can in
more » ... ple cases, such as a single-material object in air, be done with robust and quantitative methods [1]. For multi-material samples it is a challenge, especially for polychromatic laboratory sources. We have previously described theoretical aspects and performance of two multi-material methods [2]. Here we demonstrate two examples of biomedical samples that benefit from this type of more advanced phase retrieval. In practice, phase retrieval can be considered as a conversion of the edge enhancement inherent to propagation-based phase-contrast images to image contrast. A filter procedure removes an edge while increasing the contrast-to-noise ratio. For a multi-material sample several different interfaces exist and the intensity of each edge vary. Applying a simple filter to this sample will either result in remaining edge enhancement at some interface or too much removal (blurring). To avoid this, multi-material methods apply phase-retrieval filters locally in the image. To modify an image locally, 3D data is required and these methods are hence limited to tomographic imaging. Of the methods compared in [2] the newer method by Ullherr and Zabler [3] was deemed faster, simpler to use and slightly more accurate. The method is performed in five steps: 1) conventional phase retrieval on raw projection images that leaves remaining edge enhancement, 2) tomographic reconstruction, 3) segmentation of the 3D image to isolate the parts with remaining edge enhancement, 4) secondary phase retrieval on part of the 3D image, 5) merging of the 3D image. The resulting images are qualitatively accurate in the case of polychromatic radiation and quantitatively in the case of monochromatic radiation.
doi:10.1017/s1431927618014149 fatcat:ov327oleqjbwrczhlvh37pguju